chore(deps): update transformers requirement from >=4.48.0 to >=5.9.0 - #150
chore(deps): update transformers requirement from >=4.48.0 to >=5.9.0#150dependabot[bot] wants to merge 1 commit into
Conversation
dependabot
Bot
added
dependencies
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 2d09871fd2
ℹ️ About Codex in GitHub
Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".
| # AI & Machine Learning | ||
| torch>=2.0.0 | ||
| transformers>=4.48.0 | ||
| transformers>=5.6.2 |
There was a problem hiding this comment.
Bumping transformers to >=5.6.2 without raising torch>=2.0.0 leaves an incompatible dependency set: upstream 5.6.2 docs state Transformers 5.x works with PyTorch 2.4+, but this file still accepts 2.0–2.3. In environments where an older torch is already installed (or constrained), pip install -r requirements.txt can keep that version and then fail at runtime/import time in transformer-backed features.
Useful? React with 👍 / 👎.
dependabot
Bot
changed the title
2d09871 to
8f869de
Compare
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 8f869deacc
ℹ️ About Codex in GitHub
Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".
| # AI & Machine Learning | ||
| torch>=2.0.0 | ||
| transformers>=4.48.0 | ||
| transformers>=5.8.1 |
There was a problem hiding this comment.
Updating transformers to >=5.8.1 while leaving torch>=2.0.0 permits environments to resolve to Torch 2.0–2.3, but Transformers 5.x is documented/tested against PyTorch 2.4+, so this dependency set can install successfully and then fail when transformer-backed paths are imported or executed. This regression is introduced by the version bump because the previous transformers floor did not require the newer Torch baseline.
Useful? React with 👍 / 👎.
Updates the requirements on [transformers](https://github.com/huggingface/transformers) to permit the latest version. - [Release notes](https://github.com/huggingface/transformers/releases) - [Commits](huggingface/transformers@v4.48.0...v5.9.0) --- updated-dependencies: - dependency-name: transformers dependency-version: 5.6.2 dependency-type: direct:production ... Signed-off-by: dependabot[bot] <support@github.com>
dependabot
Bot
changed the title
8f869de to
de9399e
Compare
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: de9399eeb0
ℹ️ About Codex in GitHub
Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".
| # AI & Machine Learning | ||
| torch>=2.0.0 | ||
| transformers>=4.48.0 | ||
| transformers>=5.9.0 |
There was a problem hiding this comment.
Raising transformers to >=5.9.0 makes the repo incompatible with the existing maintenance flow in tools/maintenance/fix_vllm.sh (lines 4-5), which force-installs huggingface-hub<1.0; the Transformers v5 migration guide states v5 pins huggingface_hub to >=1.0.0, so running that script after this change can downgrade the environment into an invalid state and break subsequent transformers imports/startup.
Useful? React with 👍 / 👎.
|
Superseded by #413. |
Updates the requirements on transformers to permit the latest version.
Release notes
Sourced from transformers's releases.
... (truncated)
Commits
0a2757drelease v5.9.0e370a7ffix cohere2 tp_plan for releasef59ffd1Add new cohere2_moe model (#46115)4f41f26[loading] Free up tensors faster inside ConversionOps (#46110)d5dd7eeFix post processing RF-DETR (#46041)0b25f8c[serve] Support for reasoning (#45690)0df9b7fFix Gemma4 generation from inputs_embeds and per_layer_inputs (#46049)38a8b55Parakeet tdt (#44171)3428030Remove mask visualization tool frommasking_utils.py(#46066)dda0650user friendly error when loading audio from video (#45221)